Searcharxiv⌕ Search

arXiv subjects

Deze Zeng

Publications and source records attributed to Deze Zeng.

6 recordsLinked to original sources

HBF-EventLab: A Discrete-Event Simulator for Studying Host-Device Coordination in High Bandwidth Flash

High Bandwidth Flash (HBF) combines NAND flash density with high-bandwidth accelerator connectivity, offering additional memory capacity for large language model inference. However, limited hardware availability makes it difficult to evaluate how host behavior and device resource constraints jointly affect HBF request processing. We present HBF-EventLab, an open-source, transaction-level discrete-event simulator based on the Open Compute Project HBF specification. HBF-EventLab connects host request dispatch, command admission, fragmented write assembly, read dispatch, NAND execution, and completion feedback within a shared event loop. It explicitly models distinct resource lifetimes and separates configurable scheduling policies from ordering and resource constraints, enabling controlled exploration of host-device coordination. Experiments show how resource limits, address interleaving, fragment arrival patterns, and scheduling policies affect write completion and request latency. The results highlight the need to consider resource limits and scheduling policies jointly, accounting for their interactions with write assembly and read eligibility. We also integrate HBF-EventLab with real GPU execution of Qwen3-0.6B, combining profiled GPU activity durations with simulated weight-access delays to estimate time to first token and time per output token, with logits and output tokens remaining bitwise identical to those from native GPU execution. HBF-EventLab provides a configurable foundation for studying HBF mechanisms, exploring scheduling policies, and assessing how HBF access delays affect application-level timing. The source code is publicly available at https://github.com/sjtu-epcc/HBF-EventLab.

cs.AR↗

Beyond Prediction: Steering VLM Agents with Retrospective World Modeling

Equipping VLM agents with world modeling capabilities has shown strong potential for complex reasoning and long-horizon planning, while reducing the dependence of policy learning on costly real-world interactions. Existing methods mainly rely on prospective simulation to predict the consequences of candidate actions. However, this forward-only paradigm focuses on what will happen next and provides limited constraints for verifying whether an action is causally consistent with the observed state transition, which can lead to plausible-looking but physically incoherent behaviors. In this paper, we challenge the view of world modeling as only prospective prediction and introduce Retrospective World Modeling, a new agent learning paradigm that enables agents to reason backward by estimating the retrospective attribution distribution $P(\hat{a}{t}|s_t, s{t+1})$ for the action that most likely caused a given transition. Based on this capability, we formulate the Self-Consistency Reward (SCR), an intrinsic signal that measures the probabilistic consistency between the policy action and the retrospective explanation. Integrating SCR into reinforcement learning provides dense transition-level feedback and steers agents toward behaviors that are both task-effective and physically grounded. Extensive experiments across diverse agentic tasks show that our method substantially improves policy robustness and generalization over prospective-only world modeling baselines.

cs.AI↗

Learning by Neighbor-Aware Semantics, Deciding by Open-form Flows: Towards Robust Zero-Shot Skeleton Action Recognition

Recognizing unseen skeleton action categories remains highly challenging due to the absence of corresponding skeletal priors. Existing approaches generally follow an ``align-then-classify'' paradigm but face two fundamental issues, \textit{i.e.}, (i) fragile point-to-point alignment arising from imperfect semantics, and (ii) rigid classifiers restricted by static decision boundaries and coarse-grained anchors. To address these issues, we propose a novel method for zero-shot skeleton action recognition, termed \texttt{\textbf{Flora}}, which builds upon \textbf{F}lexib\textbf{L}e neighb\textbf{O}r-aware semantic attunement and open-form dist\textbf{R}ibution-aware flow cl\textbf{A}ssifier. Specifically, we flexibly attune textual semantics by incorporating neighboring inter-class contextual cues to form direction-aware regional semantics, coupled with a cross-modal geometric consistency objective that ensures stable and robust point-to-region alignment. Furthermore, we employ noise-free flow matching to bridge the modality distribution gap between semantic and skeleton latent embeddings, while a condition-free contrastive regularization enhances discriminability, leading to a distribution-aware classifier with fine-grained decision boundaries achieved through token-level velocity predictions. Extensive experiments on three benchmark datasets validate the effectiveness of our method, showing particularly impressive performance even when trained with only 10% of the seen data. Code is available at https://github.com/cseeyangchen/Flora.

cs.CV↗

Towards Fast Setup and High Throughput of GPU Serverless Computing

Integrating GPUs into serverless computing platforms is crucial for improving efficiency. However, existing solutions for GPU-enabled serverless computing platforms face two significant problems due to coarse-grained GPU management: long setup time and low function throughput. To address these issues, we propose SAGE, a GPU serverless framework with fast setup and high throughput. First, based on the data knowability of GPU function ahead of actual execution, SAGE first devises the parallelized function setup mechanism, which parallelizes the data preparation and context creation. In this way, SAGE achieves fast setup of GPU function invocations.Second, SAGE further proposes the sharing-based memory management mechanism, which shares the read-only memory and context memory across multiple invocations of the same function. The memory sharing mechanism avoids repeated data preparation and then unnecessary data-loading contention. As a consequence, the function throughput could be improved. Our experimental results show that SAGE reduces function duration by 11.3X and improves function density by 1.22X compared to the state-of-the-art serverless platform.

cs.DC↗

OTAS: An Elastic Transformer Serving System via Token Adaptation

Transformer model empowered architectures have become a pillar of cloud services that keeps reshaping our society. However, the dynamic query loads and heterogeneous user requirements severely challenge current transformer serving systems, which rely on pre-training multiple variants of a foundation model, i.e., with different sizes, to accommodate varying service demands. Unfortunately, such a mechanism is unsuitable for large transformer models due to the additional training costs and excessive I/O delay. In this paper, we introduce OTAS, the first elastic serving system specially tailored for transformer models by exploring lightweight token management. We develop a novel idea called token adaptation that adds prompting tokens to improve accuracy and removes redundant tokens to accelerate inference. To cope with fluctuating query loads and diverse user requests, we enhance OTAS with application-aware selective batching and online token adaptation. OTAS first batches incoming queries with similar service-level objectives to improve the ingress throughput. Then, to strike a tradeoff between the overhead of token increment and the potentials for accuracy improvement, OTAS adaptively adjusts the token execution strategy by solving an optimization problem. We implement and evaluate a prototype of OTAS with multiple datasets, which show that OTAS improves the system utility by at least 18.2%.

cs.DC↗

STAG: Enabling Low Latency and Low Staleness of GNN-based Services with Dynamic Graphs

Many emerging user-facing services adopt Graph Neural Networks (GNNs) to improve serving accuracy. When the graph used by a GNN model changes, representations (embedding) of nodes in the graph should be updated accordingly. However, the node representation update is too slow, resulting in either long response latency of user queries (the inference is performed after the update completes) or high staleness problem (the inference is performed based on stale data). Our in-depth analysis shows that the slow update is mainly due to neighbor explosion problem in graphs and duplicated computation. Based on such findings, we propose STAG, a GNN serving framework that enables low latency and low staleness of GNN-based services. It comprises a collaborative serving mechanism and an additivity-based incremental propagation strategy. With the collaborative serving mechanism, only part of node representations are updated during the update phase, and the final representations are calculated in the inference phase. It alleviates the neighbor explosion problem. The additivity-based incremental propagation strategy reuses intermediate data during the update phase, eliminating duplicated computation problem. Experimental results show that STAG accelerates the update phase by 1.3x~90.1x, and greatly reduces staleness time with a slight increase in response latency.

cs.LG↗